Archive for year: 2021

Artificial intelligence is arguably the most amazing innovation to date. Although all kinds of artificial intelligence products have appeared on the market, this is only the tip of the iceberg. Artificial intelligence has been integrated into our lives, even if it is still in the early stages of development, it has brought a revolutionary impact to society.

With the rapid development of artificial intelligence and the continuous enhancement of its performance, people’s pursuit of artificial intelligence has reached a level of obsession. In addition, the changes that artificial intelligence has brought to different industries have made business leaders and the public mistakenly believe that we are about to reach the peak of artificial intelligence research and maximize the potential of artificial intelligence. But what is the future of artificial intelligence? On the long road of artificial intelligence research, in order not to blindly and not get lost, we need to understand clearly the types of artificial intelligence and what capabilities existing artificial intelligence have.

The purpose of artificial intelligence is to make machines imitate human functions, so the degree to which artificial intelligence systems can replicate human abilities is used as a criterion for judging the level/type of artificial intelligence. Artificial intelligence that can perform more human-like functions and has the same level of proficiency will be regarded as a more evolved type of artificial intelligence, while artificial intelligence with limited functions and performance will be regarded as a simpler and more advanced type Lower type.

Based on this standard, artificial intelligence usually has two classification methods. One is to classify based on the similarity of the thinking of machines and humans, that is, to imitate the ability of humans to “think” or even “feel”. According to this standard classification, there are four types of artificial intelligence: reaction machines, limited memory machines, intelligent artificial intelligence and autonomous artificial intelligence.

1. reaction machine

This is the most primitive artificial intelligence system, and their capabilities are extremely limited. They mimic the human brain’s response to various stimuli. These machines have no memory capabilities. This means that they cannot “accumulate” previously acquired experience to guide current operations, that is, these machines have no “learning” ability. These machines can only be used to automatically respond to one or more sets of inputs. A typical example of a reactive artificial intelligence machine is IBM’s Deep Blue, which defeated the chess master Garry Kasparov in 1997.

2. Limited memory machine

In addition to the function of a reaction machine, a machine with limited memory (or limited memory) can also learn decisions from historical data. All current artificial intelligence systems, such as those that use deep learning, learn through a large amount of training data stored in memory, and finally form a reference model to solve future problems. For example, image recognition AI trains thousands of pictures and tags to recognize scanned objects. When the trained artificial intelligence model scans other images, it will label the new image according to the “learning experience”. As the training samples increase, the accuracy of AI recognition becomes higher and higher.

Almost all artificial intelligence applications today, from chatbots, virtual assistants and even self-driving cars, are based on artificial intelligence with limited memory.

3. Intelligent artificial intelligence

Today the first two types of artificial intelligence already exist and are widely used. Then the next stage of AI development will appear thinking consciousness. Although this is just a concept, researchers are innovating. Intelligent artificial intelligence is the higher level of artificial intelligence system. Thinking-conscious artificial intelligence can better interact with each other by identifying and understanding the needs, emotions, beliefs, and thinking processes of the other party. At present, the field of emotional intelligence has emerged, but to realize intelligent artificial intelligence, it is inseparable from the common development of related disciplines and interdisciplinary fields.

4. Autonomous artificial intelligence

This is the highest stage of artificial intelligence development. Self-aware AI, literally speaking, is an artificial intelligence that has evolved to be very similar to the human brain, and even developed autonomous consciousness. Perhaps the creation of this type of artificial intelligence, even if it does not take hundreds of years, will take decades to achieve. This will become the ultimate goal of all artificial intelligence research.

This type of artificial intelligence can not only understand and evoke the emotions of the people interacting with it, but also have its own emotions, needs, beliefs and potential desires. And this type of artificial intelligence is exactly what the doomsdayers worry about. The development of AI autonomous consciousness may promote the progress of human civilization, and may also bring disaster to society. Because once artificial intelligence has autonomous consciousness, it is very likely to easily acquire human wisdom and plan or even take over human beings by itself.

Another classification system is to divide artificial intelligence technology into artificial narrow intelligence (ANI), general artificial intelligence (AGI) and artificial super intelligence (ASI).

5. Artificial narrow intelligence

This type of artificial intelligence represents the existing artificial intelligence, including the most complex and capable artificial intelligence to date. Artificial narrow intelligence refers to an artificial intelligence system that can only use human-like functions to autonomously perform specific tasks. These machines can only do well-programmed things, so their capabilities are very limited. According to the former classification system, these systems correspond to all reactive and limited memory AIs. Even artificial intelligence that uses complex machine learning and deep learning belongs to ANI.

6. General Artificial Intelligence

General artificial intelligence refers to the ability of artificial intelligence to learn, perceive, understand and work exactly like humans. These systems can complete a variety of tasks independently, greatly reducing the time required for training. By copying various functions of human beings, artificial intelligence systems have the same capabilities as humans.

7. Artificial Super Intelligence

The development of artificial superintelligence may mark the pinnacle of artificial intelligence research, because AGI will become the most capable form of intelligence on earth to date. ASI has more powerful memory, faster data processing, analysis and decision-making capabilities. In addition to replicating the multi-faceted intelligence of humans, ASI will do everything very well (even beyond human level). The so-called “singularity” scene will appear in the development of AGI and ASI. Although the potential of artificial intelligence is very attractive, they may also threaten our survival and at least have a disruptive impact on our lifestyle.

If super artificial intelligence is really realized one day, what will our world look like? Artificial intelligence is still in its infancy, and there is still a long way to go to achieve this goal. For those who hold negative views, it is too early to worry about the emergence of “singularities”, and perhaps it will never happen-and there is still a long time to study the safety of artificial intelligence. For those who are optimistic about artificial intelligence, it can be said that “the road is long and I will search up and down”, and the future of artificial intelligence is even more exciting.

① The operation is simple, no programming

If you have not been exposed to statistical software before, SPSS is undoubtedly quite suitable. As one of the most widely used professional statistical software in the world, analysis results can be obtained only by menu operation. If you don’t have time to learn by yourself, SPSS can help you get started with data analysis in a short time.

② It has a wide range of applications.

For liberal arts majors, SPSS plays a role in many places, especially in the field of questionnaire analysis, SPSS is a unique existence. The entry of questionnaires, the reliability, validity test, T test, correlation analysis, multiple regression analysis, etc. in the questionnaire, SPSS is easy to grasp.

For science and engineering majors, SPSS can handle single-factor, multi-factor analysis of variance, multiple comparisons, and comparison of sample means involved in experimental design.

③ Powerful functions, advancing with the times

For decades, SPSS has been widely used in business intelligence, biomedicine, market research and other fields, providing users with functions such as data sorting, charting, and result display, which are flexible, clear and intuitive.

SPSS has strong compatibility and keeps pace with the times. STATA, SAS, R, Python, and databases are all available for docking. SPSS can call this function of Python, did you know?

The first problem encountered when doing research is choosing the topic. We must first distinguish the difference between the topic and the question. Often students will ask me how to do a research, and I will ask him what question he wants to study? Students will list some key words, such as education, agriculture, etc. These keywords are not strictly a question for you to research. Keywords are just topics that you are researching, and they are still far away from specific questions that you are researching. After completing the distance from topic to question, you have taken the first step of the research and can really start a research. The following three aspects should be paid attention to when selecting the topic.

First, what are you interested in

If you are not interested in a problem, it is difficult for you to make excellent research. A master student once said to me, “Mr. Uzair, you have a lot of ideas, just give me one and I will do it.” I told him, I cannot help you, before you are interested in this issue. In the process of studying, students will certainly be interested in certain questions. Sometimes you will have this kind of experience. When you see an article, you feel excited, sometimes you do not. This is the difference in interest. A person’s interests are related to his accumulation, reading and personal experience.

On the way here, colleague said something to me, and I very much agree: “Great love can have great wisdom.” How do you understand this sentence? When you are doing a research, you must prove that your research is important. How to demonstrate the importance of the research question? It is the research on this issue that can improve human society and bring welfare to humankind. Continue to ask, where does such a problem come from? It depends on whether we can transcend our personal joy and lose our attention to the future and destiny of the entire society. This is the source of interest. A good economist should have a strong sense of humanistic care and social responsibility. The starting point of a good research is a good question, which is more than half of success. In this sense, being a human being and being a learned person are the same. If you do not pay attention to issues that are important to the general public, you will not be able to do excellent knowledge; if you want to fight for fame and gain every day, you will not be able to do excellent research, because the issues you care about are not relevant to most people.

Second, you have to understand the problem

Choose the aspects that you think are important. Mathematics cannot tell you, what is important. ‘What is important’ depends on your own understanding. After determining the research direction, you should focus on smaller aspects according to your own understanding. For example, in all aspects related to the three rural issues, if you feel that the land issue is the most critical, you have already taken a step forward. Next, you think that focusing on “how to make rural laborers lose their land and obtain social security after they move to cities in the process of industrialization and urbanization” is an important research topic. If you start from the three rural issues, narrow it down to the land issue, and then narrow it down to “how to use land for security”, you have already transitioned from topic to question.

Third, why do you concern about importance?

 It is mainly reflected in two aspects: theoretically important and practically important. The best research is both. I cannot rule out some outstanding articles that are important in theory but not important in practice, or research that is important in practice but not important in theory, especially those in economics that have pioneering work in methodology, often academic and theoretical value without direct social practical significance.

The above three aspects need to be explained one by one when you do a research or write a thesis. Many of our classmates understand doing research as constructing a mathematical or quantitative model and do not pay much attention to “writing”. You have completed the math work and the measurement work. I want to remind you that maybe your research work is only 30% completed, at most 30%. Because you haven’t told us why it is important.

The current division of labor in social sciences, especially economics, is very detailed. So if you get one hundred articles, you may not understand the content of the ninety-nine articles at all, so which of the remaining ninety-nine articles do you read? When we usually read an article, we first look at the abstract. It will tell you what he has studied and what contributions he has made. Then look at the introduction to answer the previous questions in more detail. Finally, look at the conclusion and see what creative content this research has. Finally, look at the main part of the article, the theoretical and empirical models. If your abstract and introduction are not well written, when others see 500 words, they will not read your article. If you don’t pay attention to these, your research may not produce the social value it should have, so you must pay attention to it, and it may even take 70% of the time to write the introduction.

Once you find a question that you find interesting, innovative, and meaningful, then you need to judge whether it is feasible? Whether it is feasible in theory, first of all depends on whether the starting point is correct. Mathematics cannot tell you whether the starting point is right. For the study of a problem, you can use either a static model or a dynamic model. You can adopt a multi-period model or a single-period model. You can adopt a model with or without government. It depends on your understanding of this problem.

A more common problem is evidence. Maybe you have a very good idea. Is there any data to consider? Is there enough financial capacity to obtain the data to support this research? Secondly, we must consider whether the variables needed for research are measurable, at least in theory, whether someone has proposed an excessive amount of method. It is also necessary to consider whether the data sample is large enough. For example, time series data requires at least 30 observation points, but China’s reform and opening up has only 28 years since 1978. Using annual data can only be a solution. These issues must be thought beforehand.